Overall role of the miRNA pathway

The overall function of the miRNA pathway in ES cells has been evaluated in humans and mice by analysing the phenotypes of DGCR8 and Dicer mutants, respectively, as these two proteins have crucial roles in the production of mature miRNAs. In humans, DGCR8 (a double-stranded RNA-binding protein) partners with Drosha (an RNAse III enzyme) in the nucleus, and cleaves pri-miRNAs into ~70 nt pre-miRNA structures. These contain a 2 nt overhang at the 3′ end. Pre-miRNAs are then translocated into the cytoplasm by exportin 5–RanGTP and are further processed by Dicer to generate functional miRNAs¹⁴ (BOX 1). DGCR8 is exclusively involved in the miRNA pathway, whereas Dicer is involved in both miRNA and siRNA pathways. Therefore, the phenotype of Dicer mutants can be due to the absence of either miRNAs and/or siRNAs.

Dicer is essential for early mouse development, as its absence leads to embryonic lethality¹⁵. However, embryonic lethality does not seem to be due to the absence of ES cells, as Dicer-null mouse ES cells are viable, although they exhibit severe growth and differentiation defects and prolonged G1 and G0 phases of the cell cycle. It is not clear whether cell cycle regulation is required for stem cell differentiation, or whether the delayed kinetics of cell cycle progression is a cellular response to other defects that are caused by the absence of Dicer and/or miRNAs. In vitro, Dicer-null ES cells fail to express differentiation markers, such as hepatocyte nuclear factor 4A (HNF4A; which is endodermal) and brachyury, bone morphogenetic protein 4 (BMP4) and GATA1 (which are mesodermal), even after the induction of differentiation. This shows that miRNAs and/or endo-siRNAs might function in ES cell differentiation¹⁶. In reality, this differentiation defect can probably be attributed solely to the absence of miRNAs, because a related study revealed that only the profiles of miRNAs in ES cells, but not of other small RNAs, change in the absence of Dicer¹⁷. This study also revealed that one important function of miRNAs in ES cells is to regulate cell cycle progression, because almost one-half of the 110,000 miRNA molecules that have been detected are produced by 4 miRNA loci (miR-21, the miR-17–92 cluster, the miR-15b–16 cluster and the miR-290–295 cluster) that regulate cell cycle progression and oncogenesis.

Similar to the Dicer-null embryonic stem cells, DGCR8-deficient ES cells exhibit either delayed or reduced expression of differentiation markers, as well as delayed kinetics of cell cycle progression¹⁸. Most DGCR8-deficient ES cells are arrested in the G1 phase, which indicates that the main function of the miRNA pathway is to promote the ES cell cycle at the G1–S-phase transition. In addition, DGCR8-null ES cells exhibit differentiation defects as they fail to stably silence the expression of self-renewal markers, such as OCT4 (also known as OCT3 and POU5F1), REX1 (also known as ZFP42), nanog and SOX2 (REF. 18). In support of this, injected DGCR8-mutant ES cells into host mice fail to differentiate into different germ layers, in contrast with injected wild-type ES cells that can give rise to teratomas — tumours that can give rise to cell types of all three germ layers¹⁸. The similar phenotype of DGCR8 and Dicer mutants further confirms that Dicer in ES cells functions mainly in the miRNA pathway and that the predominant function of miRNAs is to regulate cell cycle progression during stem cell differentiation.

ES cells have distinct miRNA signatures

Whereas the Dicer and DGCR8 ES cell mutants provide an overall assessment of the function of the miRNA pathway in ES cells, cloning and deep sequencing of miRNAs from stem cells have revealed the identity of the specific miRNAs that are expressed in ES cells and might function in self-renewal and differentiation of ES cells. For example, in mice, the cloning and sequencing of small RNAs using conventional methods (BOX 2) revealed that the miR-290–295 cluster and miR-296 are specific to ES cells and that their levels decrease as the stem cells differentiate. In a simplified way, this indicates that the miR-290–295 cluster has specific roles in maintaining pluripotency¹³. However, more recent studies indicate that the real role of miR-290–295 is to induce differentiation (see below and FIG. 1). By contrast, levels of miR-21 and miR-22 increase substantially following the induction of differentiation, which indicates that these miRNAs might have important roles in stem cell differentiation (see TABLE 1 for a list of miRNAs that have been identified in various types of stem cell lineages). A similar study in human ES cells also shows that pluripotent stem cells have unique miRNAs, the levels of which decrease following differentiation¹⁹, even though no miRNA correlated to human ES cell differentiation has been identified. The functions of the ES cell-related miRNAs are being explored.

Box 2

DNA sequencing techniques for identifying small RNAs

Open in a separate window

The past few years have seen the emergence of new techniques for the identification of small RNAs. Cloning and sequencing (see the figure, panel a), which was the first technique developed, involves the isolation of small RNA, the ligation of 5′ and 3′ adaptors, reverse transcription followed by PCR amplification, concatemerization and cloning into plasmids. These plasmids are then transformed into bacteria and each plasmid is then sequenced.

Roche/454 sequencing (see the figure, panel b) involves adaptor ligation (both 5′ and 3′) to small RNA, reverse transcription and then capturing onto microbeads, followed by emulsion PCR. In this process, PCR is performed in an emulsion droplet that contains a single bead with a template, which results in clonal amplification of the DNA. The DNA that is attached to the beads is then sequenced in picotitre plates using the Roche GS20 sequencer, which generates up to 2 million reads of 100-nucleotide (nt)-long molecules. However, the new FLX system from 454 Life Systems and Roche Applied Science can read 200–300-nt-long molecules.

In Solexa sequencing (see the figure, panel c), adaptors (both 5′ and 3′) are initially ligated onto small RNAs and then reverse transcribed and attached onto flow cell surface, followed by amplification (dotted arrow). This results in ~1,000 copies of the same template, which are then sequenced using four-colour ‘sequencing-by-synthesis’ technology (sequencing during the polymerase reaction on a solid surface), which generates up to 40 million reads of <36 nt.

Massively parallel signature sequencing (MPSS) (see the figure, panel d) initiates with cloning a cDNA library of RNAs into plasmids, and PCR amplifying the cloned cDNAs with fluorescent primer on the 5′ end and ‘tag’ on the 3′ end. These are then annealed onto microbeads that are pre-coated with 3′ spacers and anti-tags (not shown). The small-cDNA-annealed beads are then sorted using fluorescence-activated cell sorting (FACS) and later sequenced. Since there is no PCR step involved, MPSS gives an additional advantage of measuring the expression level of small RNAs. Originally used for mRNAs, this method has been adapted for small RNAs.

Open in a separate window

Figure 1

RNA regulation of embryonic stem cells

Embryonic stem (ES) cells have the ability to self-renew or differentiate into progenitor cells, which finally give rise to all of the adult tissue stem cells in the body. MicroRNAs (miRNAs) regulate both self-renewal and differentiation pathways of ES cells by regulating various factors that mediate these processes. miRNAs form an integral network with transcription factors in regulating stem cell processes. For example, RE1-silencing transcription factor (REST), a transcriptional repressor, downregulates miR-21, which targets Nanog, SOX2 and OCT4 (also known as OCT3 and POU5F1). These are essential for stem cell self-renewal. Therefore, REST promotes self-renewal²⁶. Self-renewal is also promoted by the miR-290–295 cluster, which targets retinoblastoma like 2 (RBL2) — a repressor of DNA methyl transferases (DNMTs) — which methylate CpG islands and epigenetically silence OCT4 (REFS ^29,30). Another mechanism of promoting self-renewal is by the regulation of let-7 miRNA processing by LIN28, which directly binds to a consensus sequence in the pre-let-7 miRNA loop region and inhibits processing by Drosha and Dicer^32,33. miR-296 promotes ES cell differentiation and miR-22 inhibits ES cell differentiation. However, their mRNA targets have not been identified. Positive and negative regulations are shown by an arrow and a T line respectively; a dotted T line suggests a possible pathway that needs to be confirmed.

Whereas a few of the identified miRNAs are conserved between human and mouse ES cells (miR-296, miR-301 and miR-302), a large proportion of the identified miRNAs are unique to human ES cells. The small overlap between the two data sets might reflect the fact that human and mouse ES cells are not found at the equivalent developmental stage. In addition, the small overlap seems to be partly due to the fact that the complete repertoire of miRNAs in stem cells has not been revealed in either species, and evolutionary differences are unlikely to be a major contributing factor. In support of this, a recent miRNA sequencing effort using the Solexa sequencing technology (BOX 2) discovered several miRNAs in human ES cells that overlapped with those detected in mouse ES cells, and also with those that are differentially expressed between self-renewing and differentiating daughter cells²⁰. The development of this technology has significantly increased our ability to discover new miRNAs and other small RNAs, especially those with low expression levels or with modest differences in expression levels between various stages of stem cell differentiation²¹. It has to be noted that in conventional cloning and sequencing, miRNAs that are abundantly expressed in a cell are preferentially cloned and exclude those with low expression levels but possible significant functions in stem cells. Other deep-sequencing technologies, such as massively parallel signature sequencing²² and the Roche/454 platform^12,23,24 (BOX 2), are also ready to use to identify a plethora of small RNAs in ES cells and various other tissues.

There are two issues that need to be addressed regarding the profiling of miRNA expression in relation to ES cell self-renewal and differentiation. First, individual miRNAs can show dynamic changes during stem cell differentiation, and therefore cloning and sequencing of miRNAs before and after differentiation would only allow us to assess ultimate changes, thereby precluding the assessment of dynamic changes of miRNAs during differentiation. This could be resolved by examining the expression of miRNAs at multiple time points during the course of stem cell differentiation. Second, the varying origin of ES cell lines and the heterogeneity of differentiated cells can skew and/or mask the true expression profile of miRNAs. This might have contributed to the varying results from different searches for ES cell-specific miRNAs. The recent application of single-cell real-time PCR to profile miRNA expression in primordial germ cells²⁵ can be applied to evaluate the severity of this problem, as well as to help bypass it. In addition, homogeneous starting material and improved differentiation and cell selection protocols should also overcome the problem.

miRNAs regulate multiple steps of haematopoiesis

Haematopoiesis is initiated by self-renewing divisions of haematopoietic stem cells, which give rise to complex cell types of myeloid and lymphoid lineages (FIG. 2). A combined approach that involves assessment of miRNA expression using microarray and bioinformatic prediction of mRNA targets revealed that distinct miRNA signatures fine-tune each step of haematopoiesis. For example, miR-128 and miR-181 are expressed only in early progenitor cells and prevent the differentiation of all haematopoietic lineages³⁵. Likewise, miR-16, miR-103 and miR-107 prevent proliferation of later progenitor cells, whereas miR-221, miR-222 and miR-223 control the terminal differentiation pathways³⁵ (FIG. 2).

Open in a separate window

Figure 2

RNA regulation of haematopoietic stem cells

Haematopoiesis involves a cascade of cell proliferation and differentiation in which the long-term reconstituting haematopoietic stem cells (LT-HSCs) first give rise to short-term reconstituting HSCs (ST-HSCs), followed by multipotent progenitors (MPs). MPs differentiate into either common lymphoid progenitors (CLPs) or common myeloid progenitors (CMPs). CLPs and CMPs further differentiate to produce various cells of the bloodstream. A number of microRNAs (miRNAs) have been identified that fine-tune each step of the haematopoietic differentiation. miRNAs that function in specific steps are marked, and positive and negative regulations are illustrated by an arrow and a T line, respectively. ErP, erythroid progenitor; GMP, granulocyte–macrophage progenitor; MEP, megakaryocytic–erythroid progenitor; MkP, megakaryocyte progenitor; NK, natural killer.

The first lineage-specific decision during haematopoiesis is at the differentiation of common haematopoietic progenitor cells into common lymphoid or common myeloid progenitors. miR-223 and miR-181 are expressed less abundantly in the progenitor cells but are upregulated following differentiation into the myeloid and lymphoid lineages, respectively³⁶. Largely consistent with this finding, ectopic expression of miR-181 in undifferentiated progenitor cells increases the population of B lymphoid lineage cells but not of myeloid lineage cells.

Within the myeloid lineage, several key mechanisms drive erythropoiesis. Bioinformatics analysis has suggested that miR-150, miR-155, miR-221 and miR-222 are progressively downregulated during erythropoiesis, whereas miR-451 and miR-16 are upregulated at late stages of erythropoiesis³⁷. This implicates their function in these stages of erythropoiesis. Erythroid lineage differentiation is prevented by the repression of type I activin receptor (ALK4; also known as ACVR1B) by miR-24 (REF. 38). The growth factor activin, which, along with erythropoietin, promotes erythroid differentiation, binds to ALK4, resulting in a cascade of downstream signalling that culminates in erythroid differentiation. Another mechanism that prevents erythropoiesis involves miR-221 and miR-222, which block erythropoiesis by targeting KIT (also known as CD117), a cytokine receptor that is expressed on the surface of haematopoietic stem cells³⁹.

Monocytopoiesis is regulated by miRNA-17-5p, miR-20a and miR-106a, which target the monocytic differential transcription factor acute myeloid leukaemia 1 (AML1; also known as RUNX1)⁴⁰. AML1 promotes monocytic differentiation by activating transcription of macrophage colony-stimulating factor (M-CSF), which is essential for monocytic differentiation.

Within the lymphoid lineage, the differential expression of miR-150 regulates lineage decision between T cells and B cells⁴¹. Ectopic expression of miR-150 in the lymphoid progenitor cells predisposes them to form more T cells than B cells. This might be achieved by targeting mRNAs that are responsible for B cell maturation.

As in ES cells, lineage-specific transcription factors can indirectly determine the fate of adult stem cells by modulating the levels of lineage-specific miRNAs. For example, the haematopoietic transcription factor GATA1 is required for the expression of miR-144 and miR-451, which are specifically induced during erythroid differentiation⁴². GATA1 interacts with its cofactor FOG1 (also known as ZFPM1), and brings about transcriptional activation by binding to sequence-specific elements that are distal to the miRNA gene that produces both miR-144 and miR-451 (REF. 42). In parental erythroblast cells that lack GATA1, GATA2 occupies the site but does not activate transcription, and thereby prevents erythroid differentiation. In another study, PU.1, a transcription factor that is essential for monocytic differentiation, activates miR-424. In turn, this targets the transcription factor nuclear factor I/A (NFI-A) and induces monocytic differentiation⁴³. These observations reveal specific transcription factors as upstream regulators of the miRNA pathway in cell fate regulation.

miRNAs in myogenesis and cardiogenesis

Myogenesis is also a highly regulated proliferation and differentiation process, in which the mesodermal progenitor cells give rise to myoblasts that later withdraw from the cell cycle and fuse to form multinucleated myotubes⁴⁴. Recent reports have revealed several mechanisms that are mediated by miRNAs during myogenesis. miR-1 and miR-133 are co-transcribed and processed specifically in adult cardiac and skeletal muscle tissues, but promote myogenesis in distinct fashions⁴⁵. miR-1 indirectly upregulates myocyte enhancer factor 2C (MEF2C), an essential muscle-related transcription factor, by targeting the transcriptional co-repressor histone deacetylase 4 (HDAC4) and thereby promoting myogenesis. However, miR-133 targets serum response factor (SRF), a transcription factor that is essential for muscle proliferation and differentiation⁴⁵. In another mechanism, miR-1 and miR-206 promote myogenesis by targeting connexin 43 (Cx43; also known as Gja1) mRNA and tightly regulating its protein levels^46,47. CX43, which forms the gap junctions, is essential for normal skeletal muscle differentiation, and it is thought that gap junctions promote better intercellular transfer of the signals that are involved in myogenesis. In mouse and human ES cells, miR-1 and miR-133 both promote mesoderm formation and are potent repressors of non-muscle gene expression. However, they have opposing functions during the further differentiation of the mesodermal precursors into cardiac muscle progenitors: miR-1 promotes cardiac progenitor formation, whereas miR-133 represses the process⁴⁸. The effect of miR-1 in promoting the mesodermal differentiation of ES cells seems to be partially achieved by translational repression of the notch ligand delta-like 1 (DLL1), which in turn promotes the expression of cardiac mesoderm genes and suppresses the expression of non-mesoderm genes⁴⁸. These findings indicate that muscle-specific miRNAs reinforce the silencing of non-muscle genes during cell lineage commitment from pluripotent ES cells.

Another miRNA, miR-26a, indirectly promotes skeletal muscle differentiation by targeting the histone methyl-transferase enhancer of zeste homologue 2 (EZH2)⁴⁹. In undifferentiated mouse myoblasts, EZH2, along with the transcriptional regulators yin yang 1 (YY1) and HDAC1, epigenetically silences muscle-specific genes, and, following the induction of differentiation, EZH2 dissociates from muscle-specific genes. This paves the way for transcriptional activation of these genes⁵⁰.

Similar to ES cells and haematopoiesis, muscle-specific transcription factors and miRNAs form a network to regulate myogenesis. For example, during cardiogenesis, SRF and the co-activator myocardin bind to cis-acting elements upstream of the genes that encode miR-1-1 and miR-1-2, and increase their cellular levels in cardiac progenitor cells⁵¹. These miRNAs in turn target HAND2, a transcription factor that is involved in ventricular cardiomyocyte expansion (proliferation), and help progenitor cells to choose between proliferation and differentiation. In another case, the myogenic factors myogenin and MYOD bind to cis elements upstream of loci that encode miR-1, miR-1-1, miR-1-2, miR-133, miR-133a-1 and miR-133a-2, and regulate their expression⁵².

Regulation of myogenesis also involves transforming growth factor-β (TGFβ)-mediated downregulation of miR-24. This involves the direct binding of SMAD3 and SMAD4 to the promoter region of the miR-24 gene and regulating its expression⁵³. Reducing miR-24 expression leads to the downregulation of myogenic differentiation markers in the C2C12 myoblast cell line, whereas ectopic expression of miR-24-enhanced differentiation and partially rescued TGFβ-inhibited myogenesis⁵³. These results demonstrate a crucial role for miRNAs in modulating TGFβ-dependent inhibition of myogenesis.

miRNAs specify astrocyte versus neuronal fates

Neuronal stem cells (NSCs) have to make crucial lineage-specific decisions as to whether to self-renew or differentiate into neurons or astrocytes⁵⁴. As in other tissues, the differential expression of distinct miRNAs following the induction of neuronal differentiation dictates lineage-specific decisions. For example, miR-124 and miR-128 are specific to the neuronal lineage, whereas miR-26, miR-29 and miR-23 are preferentially and specifically expressed in astrocyte lineages respectively⁵⁵. Consistent with these findings, overexpression of miR-124 and miR-128 in the NSCs prevents astrocyte differentiation and predisposes them towards neuronal differentiation⁵⁶. miR-124 targets the 3′ UTR of small carboxy-terminal domain phosphatase 1 (SCP1; also known as CTDSP1) and induces neurogenesis⁵⁷. SCP1 is an anti-neural factor that binds to a conserved response element (RE1) and prevents the expression of neural genes in non-neural tissues⁵⁸. These studies show that miRNAs regulate the fate of neuronal stem cells by targeting key regulatory factors.

miRNAs prevent osteogenic differentiation

The function of miRNA in osteogenesis (bone formation) is now emerging. Osteoblasts (bone-forming cells) originate from multipotent mesenchymal stem cells, which differentiate into osteoblasts following induction by BMPs⁵⁹. Two recent reports have revealed that miR-125b and miR-26a prevent the differentiation of mesenchymal stem cells into osteoblasts in distinct ways. miR-125b functions by inhibiting cell proliferation through an unknown mechanism⁶⁰, whereas miR-26a prevents human adipose tissue-derived stem cells (ADSCs) from osteogenic differentiation by downregulating the translation of SMAD, a transcription factor that is required for late osteoblast differentiation⁶¹. These two studies support the role of miRNAs in maintaining the fate of osteogenic stem cells and progenitor cells.

We thank members of the Lin laboratory for their valuable comments on the manuscript. We apologize to those whose works are not cited here owing to space limitations. The stem cell work done in the Lin laboratory is supported by National Institutes of Health Grants HD33760, HD37760S1 and HD42042, the Connecticut Stem Cell Research Fund, the G. Harold and Leila Mathers Foundation and the Stem Cell Research Foundation.